People Are Asking AI About Your Organization. Here’s How to Show Up in the Answer.

Someone who cares about your cause wants to help. A year ago, they would have opened Google, typed in a few words, scanned a page of links, and clicked through to a handful of sites before deciding where to give or volunteer. Today, a growing number of them open ChatGPT or Perplexity instead and ask a plain question: “What are the best organizations working on housing in the Twin Cities?” or “Who provides free legal help for renters near me?”

The AI gives them a short, confident answer. Your organization is either in it or it isn’t. There’s no list of links to browse, no homepage to land on, no chance for your story to do the convincing. The decision starts and sometimes ends inside the answer itself.

That shift is already happening, and it’s worth understanding before you decide what, if anything, to do about it.

First, don’t panic

The headlines about AI replacing search are running well ahead of reality. Google still handles the overwhelming majority of searches, and most analysts describe what’s happening as a gradual redistribution rather than a sudden collapse. Gartner has projected that traditional search volume will drop by roughly 25 percent by this year. That’s a real change, but it’s a change unfolding over several years, not a switch that flipped last week.

So this isn’t an emergency that requires a new line item in your budget tomorrow. Be especially wary of anyone selling a “get cited by AI” package with guaranteed results. Being named by ChatGPT or appearing in a Google AI Overview is hard, and these systems tend to draw from a small set of sources they already treat as authoritative. Nobody can promise you a seat in the answer.

What’s worth doing is quieter and more durable: understand what’s changing, and make a few sound decisions that pay off no matter how fast the shift moves.

What’s actually changing

Two things are different now.

The first is where people ask their questions. ChatGPT’s weekly users roughly doubled over the past year, reaching somewhere in the range of 800 to 900 million by early 2026, according to OpenAI’s own figures. People aren’t only using these tools to write emails. They’re using them to decide which causes to support, which services to trust, and in some cases to act on those decisions directly. At a recent nonprofit fundraising symposium, one speaker described donors asking an AI assistant to go find and give to an organization doing a specific kind of research, and the assistant simply doing it. The donor never saw the organization’s website at all.

The second is what happens to your traffic. When the answer appears right in the search result, fewer people click anything. Publisher and news sites have reported organic search traffic dropping by roughly a third year over year, according to Chartbeat data covering more than 2,500 sites. Most ordinary websites haven’t seen a fall that steep, but the direction is the same one. For a mission-driven organization, it means fewer people arriving on your site to read your impact stories, learn about your programs, or find your donate button.

There’s also a strange disconnect worth knowing about. The old search-ranking playbook doesn’t translate cleanly to AI. Ahrefs found that more than a quarter of the pages ChatGPT cites most often have no organic Google visibility at all, and that only about 12 percent of the sources AI tools cite actually rank in Google’s top ten for the same question. Ranking well on Google and getting quoted by AI are turning out to be two different things.

The good news most agencies won’t mention

Here’s the part that doesn’t sell a flashy new product, which is probably why you don’t hear it much: the work that makes your website readable to AI is mostly the same work that makes it usable for people, including people with disabilities.

Large language models don’t see your site the way a visitor does. They read the raw HTML and ignore most of what loads through JavaScript. They rely on clean structure, real headings in a logical order, text that’s actually present on the page rather than assembled by a script, and labels that say what things are. If that sounds familiar, it should. It’s the same foundation a screen reader depends on.

A page built with proper semantic HTML, descriptive link text, a sensible heading structure, and content that’s served directly instead of buried in scripts is a page that both a person using assistive technology and an AI summarizing your work can actually understand. Accessibility and AI-readiness turn out to be the same discipline wearing two hats.

A couple of newer pieces sit on top of that foundation. Structured data, often called schema markup, labels your content so machines know what they’re looking at: that you’re a nonprofit, that this is an event, that this is your service area, that this is how to reach you. And llms.txt is an emerging file, a sort of robots.txt for AI, that points these systems toward the parts of your site that matter most. Both are useful. Neither does much good if the underlying pages are a mess.

What to actually do

Start by running the test yourself, because it costs nothing and it’s clarifying. Open ChatGPT, Perplexity, and Google’s AI mode, and ask the questions the people you serve would ask. The best organization for a specific need in your city. Who offers a particular service in your county. The most effective group working on your issue. Watch whether you appear, and read what the AI says about you. Plenty of organizations discover they’re missing entirely, or that the AI is confidently describing them with outdated or simply wrong information.

Then fix the foundation before chasing anything fancy. If your most important content lives in PDFs, sits behind a login, or only appears after a stack of JavaScript loads, AI can’t read it, and a meaningful share of your human visitors are struggling with it too. Get what matters into clean, accessible web pages.

Be the clearest source on your own work. Say plainly who you are, what you do, who you serve, and where. Write the way you’d explain your organization to a neighbor. The clearer your content is for someone skimming on a phone, the clearer it is for a model trying to summarize you accurately. Google’s own long-standing guidance pushes in the same direction: make content for people, not for the algorithm.

Keep your facts consistent everywhere they appear. Your website, your Google profile, your listing on Candid or Charity Navigator, your association directories. AI systems cross-reference these, and contradictions make you a less reliable source in their eyes.

And don’t pour your limited budget into trying to game the citations. Measure what you reasonably can, keep publishing honest and useful content, and let a solid foundation compound over time. That’s how trust builds in these systems, slowly and on the strength of being genuinely worth quoting.

Where this leaves you

This is less a new crisis than a new reason the old fundamentals matter. We’ve spent 29 years building websites that are accessible and clear, not because an algorithm rewarded it, but because it’s the right way to serve the communities our clients work for. It’s a little satisfying to watch that same foundation turn out to be what AI rewards now too.

If you want to know how your organization currently shows up in AI search, start with the self-test above. It’s the fastest way to see the gap between how you describe your work and how the tools your supporters use describe it for you.

And if what you find concerns you, or if your site is the kind that’s hard for anyone, human or machine, to read clearly, that’s the work we do every day. We’d be glad to take a look with you.